Our global pages
Close- Global home
- About us
- Global services/practices
- Industries/sectors
- Our people
- Events/webinars
- News and articles
- Eversheds Sutherland (International) Press Hub
- Eversheds Sutherland (US) Press Hub
- News and articles: choose a location
- Careers
- Careers with Eversheds Sutherland
- Careers: choose a location
Consequences of unpredictable code

- United Kingdom
- Technology, Media and Telecoms
01-10-2018
Writing in the Guardian, author Andrew Smith recently highlighted the increasingly pertinent problem of unpredictable computer algorithms. Specifically, Smith’s concerns lie with algorithms that are designed by software engineers to generate yet more complex algorithms - too complex even to be understood by those engineers (at least without detailed investigation). As a result, there is a risk that semi-autonomous cars are making unpredictable decisions.
The unpredictably of such decisions has already led to serious consequences (as seen in recent reports), which has given rise to a raft of ethical questions, the answers to which appear to be lagging behind the technological progress being made. Indeed, it is not only surprising how little ethical debate appears to have taken place (at least in public) but also how little regulation has been put in place to restrain the development of computer-generated algorithms to ensure that potential implications are predictable.
It strikes me that there exists a tension between the desire to create machines with capabilities beyond ourselves, whilst needing to be able to control (and predict) the actions of those machines. To resolve the issue, we could attempt to ensure that machines adhere to an ethical code, but determining what that code would consist of, or who, if anyone, is capable of devising it appears to be far more difficult to achieve than creating the sophisticated AI that we would hope to constrain.
None of this is to say that this type of technology should not be pursued; indeed, if autonomous vehicles will ultimately reduce the number of road deaths, then there is arguably a moral imperative for us to maximise their use, as soon as we are able. However, we must beware the perils of advancing technology without being able to predict its actions. As Skype’s co-founder Jaan Tallinn famously noted, “if you program a self-driving car to get from point A to point B as quickly as possible, then the passengers would arrive at point B covered in vomit and chased by police helicopters”.
This information is for guidance purposes only and should not be regarded as a substitute for taking legal advice. Please refer to the full terms and conditions on our website.
- Assignment of arbitral claims and arbitral awards: uncertain legal landscape in France
- A round-up podcast: ESG for the UK asset management industry
- Education briefing - Student accommodation: A vision for the future
- Distribution of surplus assets in a creditors’ voluntary liquidation
- UK Covid-19 Inquiry Latest update: Module 2A